关键词: 无轨爬行式焊接机器人, 同步视觉, 熔滴, 焊缝跟踪, 凹点检测

Abstract: The trackless crawling welding robot moves freely and flexibly,which leads to the low seam tracking accuracy of the active visual sensing method with advanced monitoring errors.The passive vision sensing method has no advanced monitoring error,but requires the use of an expensive molten pool camera,resulting in increased equipment costs.Aiming at these problems,a visual acquisition system composed of dual cameras was designed,and a method for measuring the welding deviation of the simultaneous collection of droplets and welds was proposed.The image of the droplet and weld were calibrated separately,and then the image information of the droplet and weld were merged into the same image coordinate system.Then,the method of pit area detection was used to locate the tip of the welding wire in the droplet image.In addition,the straight line of the weld edge was fitted using the least square method.The positional relationship between the tip of the welding wire and straight line of the weld edge was employed to calculate the current welding deviation in the established welding deviation measurement model.The experimental results showed that welding error could be controlled within 0.2mmapplying the droplet-weld synchronous vision method,which had a higher weld tracking accuracy and important application prospects.

Key words: trackless crawling welding robot, synchronous vision, molten drop, weld tracking, pit detection